Generating interpretable reinforcement learning policies using genetic programming.
Daniel HeinSteffen UdluftThomas A. RunklerPublished in: GECCO (Companion) (2019)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- markov decision process
- control policies
- state space
- reinforcement learning algorithms
- total reward
- function approximation
- reward function
- learning algorithm
- dynamic programming
- hierarchical reinforcement learning
- markov decision processes
- model free
- reinforcement learning agents
- markov decision problems
- control policy
- fitted q iteration
- partially observable markov decision processes
- long run
- decision problems
- transfer learning
- transition model
- approximate policy iteration
- policy gradient methods
- continuous state
- autonomous learning
- sufficient conditions
- control system
- decision trees
- machine learning
- neural network